How accounting continues: 25 years of building a management control infrastructure in a Swedish municipality

New Public Management (NPM) is resilient and has proved its capacity to mutate into new forms (Caffrey et al., 2019; Lapsley and Miller, 2024). NPM’s core elements, e.g competition, output focus, and trust in accounting and control technologies, have become “embedded within different government services” (Funck and Karlsson, 2020, p. 366), making post-NPM alternatives difficult to implement in practice (Funck and Karlsson, 2023). As a consequence, the main questions concerning management accounting and control in the public sector have changed – from an interest in the spread and adoption of accounting and control elements, including how they have been resisted (e.g. Blomgren, 2003; Broadbent and Laughlin, 1998; Jacobs, 2005; Kurunmäki and Miller, 2006; Lapsley, 2008), to an interest in their roles and functions as they are now established (Sjögren and Fernler, 2019) or embedded (Funck and Karlsson, 2020) in the daily practices of public sector workers.

Recent studies in this vein show how the established accounting and control elements produce multiplicity (Firtin, 2023), and how public sector organizations and practices are held together through ongoing dynamics where such multiplicity is managed (Kastberg and Lagström, 2019; Weichselberger et al., 2024; Schrøder et al., 2022). Studies show that accountants play a role in the management of such dynamics, as they become parts of management teams (Paulsson, 2012), and developed a pedagogical role (Höglund et al., 2018) where they work to make professional values cohere with economic ones (Wällstedt, 2017). It can thus be argued that the resilience of NPM is operationalized through the accounting and control elements that keep hybridized organizations and practices together (Funck and Karlsson, 2023; Wällstedt and Almqvist, 2017).

In the present paper, we take up this argument and direct the attention towards how NPM is maintained on the organizational level, through the established management accounting and control elements developed over a long time (Sjögren and Fernler, 2019). We ask the question: how does accounting continue? We suggest that the simultaneous production and handling of multiple objects (Firtin, 2023), and the ongoing processes of de- and re-hybridization (Kastberg and Lagström, 2019; Weichselberger et al., 2024; Schrøder et al., 2022) in which accounting plays a part, are important for understanding the resilience of NPM. We argue that in these processes, accounting ties practice to a broad set of programmatic ambitions (Kurunmäki and Miller, 2011; Rose and Miller, 1992; Wällstedt, 2020), making practice politically malleable – and therefore difficult to replace.

We present a study that differs slightly from much earlier work. Empirically, we follow how a single organization has built their management control system over 25 years (with the addition of a five-year follow up reflection). Theoretically, we turn to the concept of “infrastructure” (Bowker et al., 2019; Kornberger et al., 2017; Power, 2015). Studies using this concept focus on how an object such as “impact” (Power, 2015), “trust” (Kornberger et al., 2017) or “quality” (Pflueger, 2020; Reilley and Scheytt, 2019) becomes surrounded – over time – with an infrastructure that supports it in certain forms, and allows it to circulate among actors connected to that infrastructure. The reason for these empirical and theoretical moves is to allow for an analysis that (1) makes it possible to address dynamics that occur in both time and space; (2) lets us take a broader perspective regarding which actors, practices and processes that are involved; and (3) reminds us what really went on when the old public administration was replaced by NPM – all the disruptions and experimentation that occurred – compared to the more strategically oriented actions that are evident in contemporary organizations (Caffrey et al., 2019).

This paper follows how a Swedish municipality builds their management control infrastructure over 25 years – from the early 1990s to the middle of the 2010s; complemented by a revisit in 2023. The analysis builds on Power’s (2015) framework on “how accounting begins”. As such, it will show the process in which objects such as quality, competence, fairness and economy (1) start their journey as abstract policy objects – promises or dreams, but also as problems – that should be achieved by the municipal organization; (2) begin to take on more concrete forms as they become elaborated, activities are orchestrated, and an increasingly stable infrastructure is built around them; and (3) achieve multiplicity in ways that make the daily work of practitioners highly complex, leading them to engage in strategizing. This makes accounting continue: once this long and arduous process is worked through; when the infrastructure has stabilized, and the strategizing behavior is learnt, this three-step process moves much faster.

The paper aims to provide a small piece of a puzzle, and argues that the historical process of infrastructure creation may play a role in making post-NPM reform difficult in practice. We suggest that in its “youth”, NPM was introduced not only on the level of social imaginaries (Funck and Karlsson, 2023) – it was introduced through experimentation and the introduction of diverse and disruptive practical problems on the organizational level; problems that could be alleviated by an increasingly helpful cadre of management accountants and controllers promising stability once policy objects could be accounted for and reported upon. In its mature days NPM now floods practitioners with mundane problems – framed as possibilities, and connected to a diverse set of programmatic ambitions – which need to be strategically managed by politicians, managers, accountants and professionals together; something that makes it resilient and difficult to contest.

Throughout the development of public administration and management, certain aspects have remained strong: budget games and bureaucracy (Christensen, 2014; Hood and Peters, 2004; Wällstedt and Almqvist, 2015, 2017) remains from an old administration paradigm, and calculation and strategic behavior are core of today’s NPM-based models (Caffrey et al., 2019; Sjögren and Fernler, 2019; Gebreiter, 2022). Today, several of these aspects are recurringly contested (Borell, 2019; Kastberg and Lagström, 2019; Wällstedt, 2020), but they are not replaced. It is difficult to understand why, other than that contemporary public management models seem notoriously well equipped to harbor multiple ways to see, value and do things (Schrøder et al., 2022). These models can handle the multiplicity that is produced when “the object at hand becomes another object depending on how it is calculated and valuated” (Firtin, 2023, p. 162), meaning that public sector organizations hold themselves together as practitioners are able to dynamically manage “a discrepancy among existing values and between existing values and concerns that is not valued but that is difficult to ignore” (Weichselberger et al., 2024, p. 3).

In an effort to bring together the reform oriented studies that focus on the evolvement of administrative movements such as NPM and post-NPM over time (e.g. Christensen, 2014; Funck and Karlsson, 2023), and the studies that focus on the organizational level and how management accounting and control contributes to the production and management of complexity (e.g Firtin, 2023; Kastberg and Lagström, 2019; Weichselberger et al., 2024; Schrøder et al., 2022), the present study introduces the concept of infrastructure. Studies on infrastructure – much like the literature on multiplicity (e.g. Firtin, 2023) – focus on a particular object and how it becomes concrete and manageable, while retaining a more abstract or complex form. They show that every object has an abstract programmatical form – in the public sector tied to different reform agendas – as well as more concrete technological forms that defines the day-to-day operations – and that these forms exist simultaneously, creating a sort of accretion or layering of the object (Christensen, 2014; Power, 2015).

Within the infrastructure literature, Kornberger et al. (2017) focus on how the object of “trust” becomes surrounded by an infrastructure that allows it to be quantified, bringing together buyers and sellers on a market. Pflueger (2020) and Reilley and Scheytt (2019) focus on the object of “quality”. Pflueger (2020) shows how quality entered the public management discourse as an “optimistic, vague, and undefined promise” (p. 91). This promise worked like a boundary object (Star, 2010) constituting a type of interpretational flexibility that disparate actors could agree on: “quality” became the core of a wider program regarding what the public sector should produce. Both Pflueger (2020) and Reilley and Scheytt (2019) show that when trying to concretize it, quality became not only an object of consensus but also a problem – it could not live up to promises about quantification and centralized control. Instead, the control over quality – its articulation, use and production – had to be decentralized to different actors (see also Kornberger et al., 2017). This means that quality as a solution could take multiple concrete forms in practice, while retaining its promissory and abstract form upon which everyone could agree.

In order to operationalize these ideas about infrastructure in a way that allows for analyzing a development over time and space, we turn to Power (2015). He shows how an infrastructure, based on accounting, evolves by mapping its conditions of possibility on the field level and the organizational level. The progress is as follows: First, the object that should be produced is articulated in abstract forms. Power (2015) terms these forms “policy objects” because (1) they tend to be objects that are not produced in a concrete, but in an abstract, form; and (2) relate to programmatic ambitions regarding what to be achieved at a particular time (Rose and Miller, 1992). In Power’s (2015) example, the object of “impact” is supposed to be achieved. At first, the object comes to exist on a policy level. Being a policy object, it is abstract and can be featured as important and valuable in many places, and in different times, creating a general direction in which to go next. This is followed by the object being elaborated through negotiations among different actors, achieving slightly more concrete forms through “exercises” where different ways of achieving it are put on paper and discussed, whereafter some are selected as more promising than others. As a consequence, the object becomes articulated as something more specific, and regulations on how to achieve it are put in place. This makes the object thinkable and practicable (Bowker et al., 2019; Miller and Power, 2013). In the present paper, it is argued that this process can be seen as one of “programmatic object formation”, because the object becomes attached to programmatic ambitions of, e.g. “quality” (Pflueger, 2020) or “impact” (Power, 2015), brought forward by articulations in policy texts, budget documents, etc. (Kurunmäki and Miller, 2011; Rose and Miller, 1992; Wällstedt, 2020).

The programmatically formed object is then performed in a variety of ways in organizations, and after a while – often after experimentation (Andon et al., 2007) and trials and errors (Arnaboldi and Palermo, 2011) – activities aimed at producing the object in more concrete forms are orchestrated and become routinized (Power, 2015). Accounting plays an important part by defining the object and put a value on it: accounting makes what was formerly important and valuable only in an abstract sense into something that can be concretized, measured and materially created. When organizational activities are sufficiently orchestrated and routinized in order to be self-reproducing, it can be said that a management control infrastructure has been built. In the present paper this process is called “technological object formation” (Rose and Miller, 1992), thus recognizing the importance of the accounting measures, templates and IT-platforms that form and direct flows of information and knowledge among organizational actors. Figure 1 shows a modification of Power’s (2015) model.

Power’s model (2015) shows the steps through which an accounting infrastructure is created around an object, but it also shows how any object continues in its many forms: when an object has been surrounded by an infrastructure that makes it concrete, it still exists in its policy form and its elaborated form simulataneously. This is represented by the dashed and dotted squares in the model and gives us an understanding of how the concrete technological object is always connected to its most abstract form – the policy object – thereby allowing for the basic tenets of policy reform to live on. Power (2015, p. 44) argues that this type of layering (Christensen, 2014) contributes to “the accretion of elements of a transorganizational sociotechnical infrastructure”, meaning that what is arranged within the organization, through technological object elaboration, is always loosely connected to broader programmatic concerns.

The premise of the present study is the long-term engagement of the researchers with a particular municipality: the city of Stockholm, Sweden’s capital. Stockholm is a rather big organization: it grew from about 675,000 citizens in 1991 to about 935.000 in 2016. Its total income went from about 20 Billion SEK in 1991 to about 40 Billion SEK (not adjusted for inflation) in 2016, and the number of employees decreased from about 55.000 to about 40.000 during the same time. The self-image of Stockholm also changed: In 1991, the politicians in Stockholm saw the city as a local actor, at the mercy of macro-economic events, whereas in 2016, Stockholm’s brand is “Capital of Scandinavia”, and its governors saw the city as a world class, macro-economic actor of importance for the north of Europe, comparable to London and Paris.

As researchers, we have engaged with this organization extensively. Apart from a set of formal investigations carried out from 1991 to 2016 (with a follow-up in 2023), we have engaged with the city in several workshops with city officials, thesis writing students, and other interest groups during the whole work process. When compiling a study from this material, we encountered many challenges – none of which could be fully solved, but all of which could be handled. We have categorized these challenges into three main ones. We address these challenges below.

When we brought up the material we had from 25 years of studying Stockholm city, we had some ideas of what was at its core: the construction of a management control system. As such, we thought of this construction as a process that had a beginning, but not necessarily an end. What this process entailed was, however, far from straight-forward. From a historical perspective, it can be said to be a part of a long-term establishment of trust in numbers (Porter, 1996) and biopolitics (Foucault, 2008) reaching over centuries – processes that NPM is firmly embedded within. In order to keep close to the empirical material, we wanted to stay on the organizational level. This led us to the concept of infrastructure (Kornberger et al., 2017; Power, 2015): when we saw the development of the management control system as a development of an infrastructure, we could engage in the objects of accounting and management, e.g. quality, competence and economy, and analyze how the management of such objects are never closed down or finished.

To establish a field of study, we followed Chua and Mahama (2012, p. 79) who argues that “[t]he field, in our view, is not necessarily a specific locale or a firm but a constellation of theoretical ideas, empirical sites and research problems.” Instead of choosing a limited field, such as the development of the controller role (e.g. Paulsson, 2012) or the layering of reforms (e.g. Christensen, 2014) we maintained our original problem, based on the empirical material: how a new accounting infrastructure begins, but especially how it continues. This means that we attempt to broaden the field, while touching upon several narrower fields, arguing that this particular problem gives us a small piece to a grander puzzle. We see it as important for the accounting research community to be able to address problems of different scopes, where broader scopes give contributions more in the vein of discussion and dialogue, rather than a specific “deliverable” in terms of theory advancement (Chua and Mahama, 2012).

The second challenge is methodological: how to capture a phenomenon studied and analyzed over a period of 25 years, given that our empirical data is not only extensive but also drawn from five different research projects – each with slightly different purposes. The common thread among these various research projects is that they all fall under the umbrella of “Management Control” and the development of Management Control Systems (MCS). The projects shed light on the processes of infrastructures as outcomes of the introduction of competition, purchaser/provider-models, decentralization, and performance management within the organization.

Longitudinal studies are called for within management control research (Goretzki et al., 2013). We saw the opportunity to conduct a longitudinal analysis based on a 25-year perspective by analyzing the five different research projects from a unified viewpoint – the concept of infrastructure. Thus, we applied a kind of meta-analysis of the five research projects, with the building and continuance of an accounting and control infrastructure as the focal point. Our empirical data from the research projects consists of interviews, surveys and, and participative observations conducted over these 25 years with managers at various levels and employees. We depict the empirical material in Table 1.

A significant challenge we encountered was structuring a vast corpus of empirical data. The most logical approach forward was to allow a document analysis guide us in organizing the empirical findings. Therefore, we analyzed all city budgets and annual reports from 1991 to 2016, i.e. 25 years. In these documents, the political ambitions for the organization are expressed, and governance is based on them. We then had data – interviews, observations, surveys, and workshop notes – on both the work of accountants and professionals, and decided to focus on how they, in their practices, became invested in building the infrastructure. The semi-structured interviews were conducted with managers and employees in the organization. Observations stem from budget meetings between managers and controllers, as well as staff meetings between managers and employees. Our surveys were targeted toward employees within the elder care sector on two occasions. The workshops involved the presentation and discussions of our research findings with CFOs and controllers. When it comes to professionals, most of our data is on elder care which will, therefore, serve as the example of professional engagement.

We used the reading of the budgets and annual reports to structure the material and coded the material according to a five-step process (c.f. Thomasson, 2018; see also Hyndman et al., 2014; Polzer et al., 2017). During the first step, being interested in the programmatic object formation, the control technologies used, and the accretion of infrastructure, we looked for different objects and their articulations in different forms in the budgets and annual reports. During the second step, we found that, over time, there were four overarching and abstract objects that were recurring in the documents: Quality, Economy, Fairness and Equality, and Competence and Staff. In the third step, we analyzed how these objects were articulated in more specified forms, and how infrastructural elements were added. In the fourth step, we engaged with the data from the research projects, to understand how politicians, accountants and professionals had engaged with these four abstract objects over time and in their respective practices. This was done to understand the experimentation within the organization, i.e. the technological object formation. Finally, the fifth step consisted of data validation through a workshop in which we presented the results of our study and the model of analysis for the organization’s CFO, management accountants and controllers. Through this approach, we balance the strength of the meta-analysis with our long-standing empirical engagement that “enable researchers to be present for the changes as those changes were occurring” (Golden-Biddle and Locke, 1993, p. 421).

From the empirical data, we derived three phases – periods – of building the infrastructure. We revisited the organization with a new project in 2023, establishing that we are still in the third period. We see these periods as struggles to gain knowledge about what the organization should provide, and an ongoing engagement to build an infrastructure that would support the delivery of the identified objects Quality, Economy, Fairness and Equality and Competence and Staff. The first two periods, we argue, follow Power’s (2015) framework rather well; especially if we account for the development of programmatic object formation and the development of templates that mediates and governs activity orchestration and technological object formation. The third period, however, can be used to understand how the infrastructure allow the accounting-based control system to continue. We have termed the periods as follows.

1. “the period of polarization – programmatic object formations and experimentation” between 1991 and 2000;

2. “the period of control – technological object formations” between 2000 and 2010; and

3. “the period of holistic management, contradictions and strategizing – dealing with object multiplicity” between 2010 and 2016.

In order to show how we coded the data, we will end each period with a table that shows the infrastructural elements that had been built around the specific objects during that specific period, and with regards to the different organizational practices of politicians, accountants and professionals. We will show the new elements in bold in the tables, while showing those that remain from earlier periods in regular font.

When writing up the study, we took Latour’s assertion that “[t]he text, in our discipline, is not a story, not a nice story. Rather, it’s the functional equivalent of a laboratory. It’s a place for trials, experiments and simulations” (2005: 149) seriously. We have written it up in several versions before: as a performative problem (Fırtın and Karlsson, 2020; Yu and Huber, 2023), a problem of valuation (Schrøder et al., 2022) and accountingization (Sjögren and Fernler, 2019). These efforts of trials and simulations led us too deep into theoretical problems that tended to obscure our main point about how accounting continues, and what this may mean for public sector accounting. We tested our writing on seminars, and in workshops with practitioners as described above, seeing the need for a framework that was easier to grasp, while still able to showcase “the unpredictable, nonlinear and socially-embedded nature of management accounting change” (Baxter and Chua, 2003, p. 112).

This made Power’s (2015) framework even more appealing, as it provided a rather simple frame, allowing us to engage in the complexities that regards the three different practices – political, accounting and professional – and the four types of objects that we extracted from our material. In another appeal to clarity, we have also taken control over the authorship (Baxter and Chua, 2008) by using a perhaps overly objectivist language in the empirical section (deVilliers and Dumay, 2013) at the expense of authenticity (Golden-Biddle and Locke, 1993). In the end, we argue, the laboratory experiment has to stop and a coherent story be constructed, and these are the decisions we made along the way.

It is important to emphasize that Stockholm did not start from scratch in 1991. Budgetary processes were already present within the organization prior to this time. What occurs within the organization in the beginning of the first period is what previous research has termed a paradigm shift in the approach to governance in the public sector – from an old administration model towards NPM (Almqvist, 2001; Burns et al., 1994). The starting point of this shift is in autumn 1991, when politicians from the liberal-conservative coalition won the election. We have been able to identify the first clear emphasis of such a paradigm shift in the documents, beginning with the overall budget of 1992.

The start of this period was affected by a downturn in the economy. The budget of 1991, constructed by the left majority, focused on this. This budget was built on business as usual and based on traditional budgetary resource allocations, but also concerned with worries about lower tax revenues and higher costs as a consequence of macro-economic events. Traditional cost-cutting measures, and usual projects, such as competence development of staff, were carried out throughout the period. The budget years of 1992–1994, the liberal-conservatives took over. They began by exposing elder care provider units to competition by putting 20% of the in-house units out to tender every year. Profit responsibilities were assigned to the units to make them prepare for the possibilities to be exposed for a bid. The bids could then either be won by the staff at the unit themselves, or by companies bidding for the take-over of a provider unit. The bid must include quality criteria, forming the basis for a contract and follow up. To this came a reformation of the organizational structure: a purchaser and provider split was carried out for competition to work well. This was then complemented with a system of customer choice.

When the left took over between 1995 and 1998, they saw the competition project as highly ideological. They also saw the liberal-conservative majority’s competition scheme as a financial project, aiming to cut costs at the expense of service quality. The left majority did, however, not completely disband competition. Instead, they made efforts to include more quality aspects in the scheme. They also embarked on their own pet project: the political decentralization of the organization into district committees and corresponding administrations. The argument was to make the organization more democratic, and make the political practice come closer to the citizens. The project was built on a high degree of autonomy for the new committees, and every district administration was allowed to organize themselves as they liked. It was assumed that this would boost both quality development and democracy.

At the end of the period, the liberal-consevatives took over again. Highly critical towards the left’s policies – the slowing down of privatization and competition and the focus on costly political decentralization projects – the new majority argued for more control. They continued their privatization and competition scheme, while working on the control system based on purchaser and provider differentiation and management by objectives. Instead of retracting the decentralization performed by the left majority, the liberal-conservative majority began to introduce another type of control over the organization that would come to dominate the next period. In the end of the period, the organization was structured as in Figure 2.

Accountants, during this time, were focused on traditional budgeting and cost management. A few radical changes were made. Units exposed to competition were reorganized as profit centers, and they got the possibility to transfer surplus to the next budget year. Financial indicators, regarding costs, revenues, investments and asset management were developed to retain better control over finances. The city introduced a whole of government approach to their accounting, and started to account for their finances in a consolidated statement, integrating the financial operations of the municipality and its owned corporations in the same report. Monthly budget meetings started to appear.

Competition, the purchaser provider split and the decentralization affected accountants. First, there was an increase in cost management for the elder care units exposed to competition, and second, a pay for performance resource allocation scheme, based on volume, was developed for Stockholm’s pre-schools. Somewhat counter intuitive, these two actions were separated: the elder care units exposed to competition still got their resource allocation as lump sums and as part of the traditional budget process, whereas the pre-schools, now included in a customer choice system, would get their revenues depending on how many customers they attracted. To support managers with new profit center responsibilities, projects to enhance financial administrator competence were launched.

Financial matters were, hence, at the core of accountants’ practices. Matters of quality tended to be neglected – it was still an abstract object; a dream or promise. Although one of the ideas of the competition program was to include quality criteria in tenders and contracts, this was not working well – the only value that was regularly included in the tender and contracting practices was staff and management competence.

The last few years of the period, there were efforts to make the different organizational units formulate commitments (what they committed themselves to achieve in terms of service quality) and corresponding work procedures (how they were to achieve these commitments) and relate them to the city’s overarching goals. The ambition was to measure these commitments, but this was never implemented during this period. Instead, several groups were formed, doing exercises aimed at coming up with suggestions for reasonable quality measures in different areas.

In the beginning of this period, the responsibility for much of Swedish elder care was transferred from county administrations to primary municipalities like Stockholm. At the same time, political practice championed more individualized services. Routines were disrupted by the two main projects launched during this period: the exposure of provider units to competition by the liberal-conservative majority, and the subsequent decentralization of the organization to district administrations brought forward by the left majority.

The project about competition was launched early in the period. It was built upon the new profit center responsibility, and the lingering threat that every provider unit may be privatized and exposed to competition. Because financial resources were still allocated as lump sums to each unit, within elder care, the only opportunity to come out on top was better cost management, making costs the main economic objects to be handled. Most of the unit managers we interviewed had worked hard with exercises going through their operations and looking into what drove costs. They reported cost savings, but also a fair amount of stress from the pressure to reform their operations.

The decentralization project brought about confusion. Especially in elder care, which had gone through several changes in a short time: first it was moved from being a responsibility of the county, then it was exposed to competition, and now it was supposed to go from a centralized structure to a decentralized one. To this came the opportunity for the districts to organize their operations as they wanted, which furthered a division of the organization: although this enabled each district administration to develop their own practices, there were few, if any, communication over administration boundaries. Everyone had to manage themselves, giving the professional practice the opportunity to develop specialized qualitative services, but also making it difficult for practices to learn from each other, over functional boundaries.

In summary, it can be observed that this period is characterized by policy object formation in the form of articulating NPM-oriented ideas and concepts. This is still at the idea stage coupled with experimentation and trial and error on the operational levels: object elaboration begins, during which experimentation starts to take shape; practitioners are exposed to new problems, and begin to experiment and test. As we will see, all of this will settle and stabilize in the next period. Table 2 summarizes the infrastructureal elements that were built around the four objects this period, and with regards to each role.

The period of control was, from the political side, saturated with rhetoric of results focus, quality guarantees and individualized services. Such practices were to be carried out by empowered, invested and competent co-workers, as well as clients empowered by enhanced freedom of choice and more involvement in decisions regarding their received services. Moreover, there was a strong focus on routinization regarding follow ups: the use of interim reports and forecasting was enforced. At the same time, we could see remnants from the period of polarization: where the liberal-conservative majority that ruled during the first two years of the period of control continued with a tough competition and privatization regime, the left followed up with more rigorous needs assessment routines, aiming for more fairness in resource allocations, and a focus on competence development. The left majority used the income (2 Billion SEK) from the sale of the municipality’s energy company – an achievement by the privatization-focused liberal-conservative majority in the last year of their tenure – to establish a competence fund. This project brought together co-workers from all of the different professional areas in the city and invited in-house service providers to a new exercise: writing applications for their own competence development projects. In sum, around 500 local projects were initiated.

During the whole period, the entire organization was working with a new management control system, called “integrated system for management and control” (ILS) based on objectives and corresponding work procedures. This way to orchestrate activities around articulated objects was typical during NPM: the politicians articulated “what” should be achieved (objectives) whereas the professionals were to articulate “how” these things should be achieved (work procedures). There was also an opportunity for co-workers to re-formulate objectives and turn them into objectives and “quality guarantees” fitting local activities. Abstract objects, articulated by politicians, and locally articulated versions of these objects, related to specific professional practices, could be linked to work procedures and become the basis for quality development. Empowerment, dialogue and recurring follow ups through interim reports every four months were part of the scheme, allowing for the production and disclosure of high-performing service delivery.

The period of control meant a new role for the accountants and a transformation of the accounting practice: from being “bean counters” focused on costs and balanced budget, they now had to develop skills in all sorts of areas. The accountants became invested in the development of the management control system. The system was designed to handle all types of services, and both in-house and private service providers. It stipulated general objectives (normally three or four) broken down to operational objectives (from 17 to over 30, depending on political majority) on the council level. These objectives were then broken down on committee level, and the system then allowed providers to reformulate these objectives into articulations fitting their own activities. These articulations were called “commitments” and “quality guarantees”. Accountants governed this work, but the articulation exercises were conducted by unit staff and managers on designated staff meetings.

The design, as it was developed towards the end of the period, and continued into the next period, is depicted in Figure 3. We include an example of how a general objective was broken down and reformulated on the different levels. It should be noted that the particular general objective used in the figure was reformulated into several operational objectives, objectives on committee level and commitments and quality guarantees on the unit level, whereas we only show one example of each.

Accountants established a role as business consultants for unit managers where they tried to complement managers’ focus on costs with more focus on revenues. Later in the period, they became involved in the development and interpretations of quality indicators, conducted follow ups regarding how needs assessments practices allocated resources among districts, and participated in monthly budget meetings with unit managers. They tried to work as “benchmarkers”, looking at units achieving high levels on indicators, urging “low achievers” to get inspiration from how these “high achievers” articulated their commitments and work procedures. In short, accountants had to come closer to the professionals in all these roles to help them disclose their work practices.

Nevertheless, the accountants were still mostly preoccupied with financial matters: economy was the main object. They constructed standard cost calculations that could be used as transfer prices in the purchaser-provider model. Costs were followed up for all units – both purchaser and in-house provider units, and every unit manager had monthly budget meetings with “their own controller”. Forecasting became important, meaning that the accountant needed information from the managers. Cooperation and mutual dependence was developed: monthly budget meetings became as much a control endeavour as a possibility for both controllers and managers to brainstorm about possibilities and obstacles in the pursuit of more effective value creation – both parties became invested in turning specified problems into solutions. Cost management was seen as the main problem, but by introducing a new solution – revenue management – managers could become more invested.

The monthly budget meetings were complemented by a stronger focus on interim reporting during this period, as part of the ambition for better forecasting and, consequently, better budget management. In the end of the period, quantitative indicators were introduced in both the annual report and interim reports and started to form the bulk of how service quality was disclosed and evaluated. In 2007, the forecasted fulfillment of objectives was described almost only in text, and by signaling them in colors: green for “high probability for achievement”, yellow for “will probably be partly achieved”, and red for “will probably not be achieved”. In 2009, the colors remained, but were now based on quantitative indicators and how their levels corresponded to budgeted target levels. Table 3 gives an example of indicators used for nursing homes, between 2008 and 2016 in budget documents and annual reports.

From 2008, all budgets, interim and annual reports were based on quantitative indicators. Also the individual units’ operational plans and their follow up documents were connected to the system of quantitative service quality evaluation. This was significantly helped by the implementation of a web-based system, which offered a template for input of data, and the subsequent possibility to extract data in the form of standardized reports: with the web-based system, everyone in the organization could view the articulated objectives and commitments on different levels, the development of the quantitative indicators, and the formulations about work processes in each unit. It was a system of disclosure that opened up different practices to each other in particular ways; an infrastructure assembled around objects, in quantitative forms, that connected the political practice, the accounting practice and the professional practice.

Much of this period, in the professional practice, was characterized by the work with the competence fund during 2002–2006, the simultaneous implementation of the management control system, and an ambition to develop the control system to ascertain that service production was fair and equitable over the city. There was a continued ambition to secure a cost focused organization that would keep the budget balanced. Staff meetings came to center on the articulation of objectives and quality guarantees: professionals now had to engage in exercises where they determined what levels on every indicator they saw as possible to attain during the budget period. In this period, the professionals used a depiction as in Figure 4 on how to operationalize the stipulations of the control system. This was, then incorporated into a reporting system where every unit had to compile more detailed four-monthly and year-end reports, specifying their results and how and why they managed to (not) achieve them.

The competence fund started in 2002, and was a huge project incorporating almost the whole organization. The idea was to fund a multitude of competence development projects, where units, work teams and others could apply for money. The scheme was fraught with uncertainty from the beginning. Because “competence” was abstract – it had been talked about and slightly elaborated, but seldom worked with – so no-one really knew what it was: there was a need to formalize what could be understood as competence. The demands on the applications became tougher, and low cost appeared as an important value in the application. Cost effectiveness became one of the most important criteria in the application writing practice.

Competence enhancements that could be easily measured became emphasized, much because the competence fund financed the salary for co-workers being sent to training courses. The value of low costs became tied up with the possibility to produce competence as an object that would “fit” into the formalized system. This was also a consequence of political ambitions. Formalization was hence important and led to an increase in documentation on all levels. An important issue was for managers to write applications for competence development during the implementation of the competence fund. These applications were then to be followed up with thorough documentation regarding what was being achieved in terms of costs and capabilities to produce more qualitative services. These exercises were not total news for the managers, because they got some training in documenting what was going on in their organizations when they were exposed to competition during the former period. Such documentation could then be used by managers meeting each other in work groups, over organizational boundaries.

This also fueled the simultaneous work with formulating objectives, quality guarantees and work procedures according to the templates of the management control system: professional activities became orchestrated around more concrete objects, and, consequently, more formalized and routinized. Managers and co-workers from different areas became mutually invested in disclosing their work in forms that could easily be retrieved by others. A vague rhetoric of dialogue and involvement, coupled with substantial investments brought forward through the competence fund, made cooperation over boundaries possible: the ideas of dialogue and involvement formed an environment where objects in concrete forms could be discussed, whereas the competence fund formed the basis for the cooperation that orchestrated activities throughout the organization.

In summary, it can be noted that experimentation continued during this period but also evolved more towards a focus on technological object formation and activity orchestration. For example, experimentation persisted when local in-house units were encouraged to submit project applications for funding for competence development from the Competence Fund. This evolved into a platform for collaboration between units orchestrating activities across the entire organization. The emergence and concretization of broken-down indicators at various organizational levels are also examples of technological object formation (see Table 4).

In this period, the political ambitions grew – much because there seemed to be an understanding that sufficient control of the organization was achieved. Politicians therefore started to look both more outside the organization, and further into the future. We could see this already in the end of the period of control, but in this period the political practice went from tactical to strategic. Part of this also stemmed from a sense of changing complex environments: when the organization appeared stable, it was the surrounding world that needed to be dealt with.

The liberal-conservative majority in charge in the beginning and most of the period worked hard with slogans. Their most important one was “a world class Stockholm”. The view of Stockholm as an important international and macro-economic player was taking hold among politicians, and the idea was that even if Stockholm was good, it still held more potential. It was therefore important to continue working with what was already working well – for example budget management and leaders’ ability to enhance employee commitment – and to improve what did not – for example the competence of those working in the city’s purchaser functions. An opportunity to be of assistance to the surrounding society, while making use of and coordinate the resources in it, was identified.

As a consequence, politicians articulated overarching ambitions: the organization needed to be flexible in all kinds of work, and multiple providers should be available so that citizens could find services that suited their particular needs. For the liberal-conservative majority, it was predominantly the market that would provide flexibility and multiplicity, whereas the left majority, coming to power the last two years of the period, still saw the city’s organization as the more important player that should, in itself, become more flexible and varied. These differences were, however, considerably smaller than in former periods and little was really changed when political majority shifted.

Investment became a new focus – partly as a long-term consequence of the financial crisis 2007/2008 where the city could use its strong balance sheet to boost the economy and, in return, get cheap investments in for example property and infrastructure. At the same time, there was an ambition to keep working with costs, and specifically lower overhead costs: the rhetoric was that the funds should go to the operations and not to administration, energy bills or rent. The ambition to lower energy costs cohered well with the focus on investment and collaboration: by participating in the formation of research projects, start-ups and other things related to environmentally friendly technologies, Stockholm could advance several of its agendas simultaneously.

For politicians, Stockholm was articulated as a whole; a space for everyone. Therefore, accessibility became an important issue – both in terms of accessibility to places and in terms of democracy. The web-based digital systems were developed further in order to include more services and more citizens. Moreover, there was an enhanced interest in working with gender equality – an ambition we had not seen clearly before. Overall, the political ambitions during this period were directed “outside” the organization, rather than to the “inside”. As a consequence, almost everything politically conceivable came to be included as potential objects to steer toward.

Accountants continued to work on things implemented in the former period. Much had to do with enhancing their competence, both in more technical terms, such as developing costing models and spreadsheets for follow-ups, and in terms of communication and pedagogy. In order to function as business support for managers, many of the accountants we interviewed had worked to improve their pedagogical skills. They also used this to educate both managers and co-workers in business management. Several of our respondents had, if they were controllers, held classes or, if they were professionals, attended such classes. Forecasting, and dealing with all sorts of accounting information, were seen, in this period, as a fundamental competence for unit managers, and, to some extent, also for the individual professional. Financial management had become normal.

Most managers had, at the end of the former period, learned to work not only with both costs and revenues. The work with revenues for managers of nursing homes in the elder care became quite specific, where needs assessments were formalized into categories coupled with remuneration levels: the “higher” the needs of a user, the higher the financial compensation. The assessment practice, combined with the remuneration system of pay for performance, made it possible for accountants and the professionals at the provider units, to mutually invest themselves in the maximization of revenues by engaging in strategizing over what constituted individual needs. We will discuss this from a professional point of view below, but from an accounting perspective the system with assessments and corresponding remuneration levels sometimes had grave consequences. If a higher level client passed away, and was replaced with a lower level client, the drop in revenues could mean that an employee had to be let go. There was no way to control which “level” the new clients would be, because of the freedom of choice.

Freedom of choice was hence a problem for the providers when combined with other things that had become concretized during the building of the control infrastructure. Not only could they get clients with “wrong” levels of needs and corresponding remunerations, given how they organized their service provision; they also had to attract clients. The most important way to do this was to take control over what signified quality – have well-articulated quality guarantees, and high scores on the quantitative indicators. The indicators were not only used internally, as a way for politicians to articulate the commission the professionals had to adhere to, they were also displayed on the city’s website so they could be used by prospective clients to benchmark, and compare different nursing homes with each other. These measures were, hence, doubly important.

During this period, also professional practitioners seem to have learned to work strategically within an infrastructure being holistic in terms of the overarching and abstract objectives and its allusion to empowerment and dialogue but, simultaneously, being quite detailed and concrete in terms of the introduction of quantified indicators, and how co-workers were supposed to account for and formalize their commitments and work procedures.

Management and leadership were seen as important parts of producing the desired objects. There was a struggle over what should be known as “good practice”: there was an important idea about “close leadership”, that managers should be close to the professional practice, but “management by numbers” through the formalized, web-based, control system mattered as much – often more. High levels on the indicators was understood by top management and politicians as having produced an object such as quality, but close relationships and dialogue were what made things work, argued the interviewed managers. Results in the form of quantified indicators, together with its connection to working procedures as stipulated in the routinized reporting system (see Figure 4) were now normalized and accepted as a common truth within the organization. By connecting “best practice” to articulated commissions – what, in terms of, e.g. quality, was supposed to be delivered – it was assumed that objects could be produced more efficiently. There were, however, not full consensus regarding what the commission was. A district manager:

“I think that, by working differently, we can get 15–20% more out of every SEK [the Swedish currency]. [We can achieve this] partly by talking a lot about ‘what is the commission’ … because there are still many [co-workers and unit managers] that tend to, well not formulate their own commission, but almost … And that leads to them doing things outside the commission, and that is very expensive”.

There was, thus, a still a struggle over in what form objects should be produced: there was a multiplicity of possible versions to attend to; and according to this manager, value creation could be made much more efficient if that struggle would cease. Only one version of any object such as quality would be the right one. This did not work well for professionals. An elder care manager, working closely with her staff said:

“Every year the politicians tell us that ‘now you should be working with these five things, because we see signs that these are things you can develop’. ‘OK, these five things,’ we say. And then we look where we stand on those things and ask questions [to staff and clients] about them. But then there is so much more that is our responsibility. If the politicians say that it is important that you get good meals, that it is important that relatives feel secure … And then the individual comes up with something completely different: ‘You know, for me it is important that my apartment is clean when people come to visit, that would make me satisfied’. That is not captured by the control system. But that doesn’t mean that we shouldn’t do it.”

Working at the intersection between individual needs and stipulations of the control system, this manager had to deal with dissension: not only what was captured by numbers in the control system could be understood as the desired form of a particular object. There could be dissension about exactly what constituted certain kinds of “quality”, and how to orchestrate activities around them.

Such dissension involved documentation. Because of the design of the remuneration system, the formalization of the needs assessment process, and accountants’ willingness to educate managers and co-workers in financial issues, documentation became a struggle between different objects: depending on if it was used to maximize revenues, or to enhance the quality of individualized work procedures, the documentation related different items of the infrastructure to each other. When documentation became orchestrated around the concrete economic object of revenues, the focus was on documenting deviations from the standards, aiming to “level up” users with “too low” needs assessments, thereby increasing revenue. When orchestrated around the object of quality, and especially individualized services, documentation became something else – a way to concretize the fluid object of quality into work practices.

Thus, many of the managers and co-workers in our studied administrations and units tried to “strategize” the system: they worked with their own solutions regarding everything from documentation to managing quality indicators. In order for professionals to produce what was desired, they had to orchestrate activities around concrete, articulated and measurable objects and construct routines. But because professionals work with multiple objects in dissension, these routines had to be constantly challenged and developed. The control system where provider units disclose their work, and the results they achieve, provides possibilities to benchmark and learn from each other. Such best practices, however, have to be adapted to the teams and specific conditions of the local: each unit has to make up and develop their own routines. The underlying understanding of the professionals – managers as well as co-workers – seemed to be that for the infrastructure to support multiple dissenting objects, it had to be flexibly managed: the system had to be strategically exploited in the short term, and developed in the long term.

In summary, it can be observed that this period begins with an expansion of the organization’s perspective. Stockholm was depicted as a whole; a space for all, demanding a holistic management within the developed infrastructure. This naturally has implications for the management control model, which must now capture and form additional objectives and subsequently orchestrate activities and routines in a flexible and varied manner to satisfy various stakeholder needs. Here, the ability to strategize the system becomes crucial. Professionals within the organization learn to manage multiple objectives, such as “leveling up” clients to attract revenues, or developing quality documentation in such a way that aspects perceived as important for quality, but not explicitly defined as such in a programmatical sense, are still implemented. Strategizing the system became a means to satisfy both the goals and requirements of the system, as well as the professionals’ and customers perception of, e.g. “good quality”. The call to arms was: “Flexibility! All the time!” Table 5 summarizes the infrastructural elements that were built around the four objects this period, and with regards to each role.

In 2023 the infrastructure remained stable: if there were new political goals, they were broken down in sub-goals, performance indicators were constructed, and activities articulated. In the spirit of keeping up with Stockholm being a global player, the overarching goals have become more connected to Agenda 2030 and its 17 global sustainability goals. One development was that the performance indicators were no longer considered drivers of change, instead the goal was to keep them on a decent level. Instead, there were more focus on activities that had been articulated on the political levels. These activities would be regarded as projects to be carried out during a limited period, and then incorporated into the ordinary work. Some projects are more concrete, such as projects to decrease the number of casual staff, or to change to a digital lock system in domestic care, whereas others are more strategic such as developing collaboration with other districts or other societal actors.

Even better conditions for strategizing had been developed. It was quite clear that one important factor for good strategizing was to have an optimal size: if a district is too small, it would have less room for moving things around and de- and reprioritize things; and if it is too big, it would be difficult to have a holistic view. Therefore, several smaller districts had been merged. In order to be able to strategize, there also had to be a flux between stability and flexibility: sometimes the work with an object such as quality belonged within a certain unit, and sometimes it had to be the object of collaboration between many units, where organizational boundaries did not matter. This was also the case for economy. The district budget needs to work for the whole district, as a unit, but also for the units within the district. Collaboration and cooperation were thus emphasized.

This paper describes how a management control infrastructure was built in a municipal organization over 25 years. It does so by invoking the concept of “infrastructure” (Kornberger et al., 2017; Power, 2015). More specifically, it follows Power’s (2015) model on how accounting begins. In this section, we will begin by showing exactly this – how accounting begins – because it serves to remind us that apart from traditional budgeting and cost accounting (Wällstedt and Almqvist, 2017), there were few accounting objects present before NPM, and that there was little experience on how to manage and organize around them. We then continue by developing the model to discuss how accounting continues, and why it might be difficult to replace NPM as long as accounting persists in this way.

In the empirical material, we could see a phase of polarization and experimentation where quality, economy, fairness and competence were created as policy objects. There was a political struggle to define them, and there were connections to broad reform ideas such as competition and purchaser-provider splits from the right, and a focus on decentralization from the left. Apart from classical conceptions of costs, the objects tended to be abstract and although there was some consensus that, e.g. quality was important, there were different ideas on how to achieve it. A high degree of experimentation (Andon et al., 2007) and trial and error (Arnaboldi and Palermo, 2011) could be observed, here, when building the infrastructure. Under the pressure of the different projects – competition and the division of the organization into purchasers and providers; and the decentralization project into district administrations – accountants and professionals were engaged with working within their own enclosures; keeping to their own traditional ways, while trying to make sense of contradictory signals from the polarized political level.

To break through these enclosures, a set of problems, connected to reform agendas, were introduced. These problems were created by the politicians themselves – probably as much by chance as by design, given the political polarization. Through object elaboration (discussing what constituted, e.g. competence and quality) and recurring documentation exercises, based on cost effectiveness and competitiveness, concrete problems were formed for professionals. But as they were allowed to address the problems and generate their own solutions, the programmatically formed objects emerged as the new normal for all actors working within the organization: professionals who became exposed to problems were empowered to create solutions themselves, leading to both problematizations and solutions becoming “their own”. It was through these processes of object elaboration that the policy objects became better understood: it became possible to think about them more clearly (Bowker et al., 2019; Miller and Power, 2013), and devise more specified articulations regarding their production (Kurunmäki and Miller, 2011).

After this, a pursuit to achieve technological object formations commenced – a pursuit that involved the orchestration of activities around problems and solutions that were becoming more and more concrete. An apparent problem was the need for a clear management structure for professionals and their managers. This was not necessarily the first priority for professionals, and they could probably resist (Broadbent and Laughlin, 1998; Kurunmäki and Miller, 2006; Lapsley, 2008) but were instead enrolled even more to contribute themselves to develop the infrastructure. Both through the pressure of competition, but also by being invited to suggest what would count as, e.g. “quality” from their perspective, and by being selected for competence development with the possibility to learn from each other. Accountants, being at work designing much of the core infrastructure, were eager to help: they both constructed the calculations that formed the basis for revenue management, and introduced the professionals to them.

There were, thus, abundant opportunities to become invested in solving problems and contribute to the production of the desired objects: these opportunities were the main vehicle of activity orchestration. The competence fund put professionals in touch over professional and organizational boundaries, and orchestrated activities around objects of costs, quality and competence. Accountants worked as pedagogues (Höglund et al., 2018; Paulsson, 2012), helping professionals to decipher the intricacies of running a profit center (Wällstedt, 2017), orchestrating activities around costs and revenues, with the consequence that professionals became their own accountants. In short, the infrastructure provided avenues where rather specific problems could be solved both by individual engagement and collaboration between accountants and professionals, as well as over organizational boundaries.

Through repetition and hard work, an infrastructure where concrete measurable values were at the core, became stabilized (Power, 2015). Objects were specified in measurable terms, and work procedures on how to achieve them – how activities should be orchestrated – were spelled out. The abstract form of the objects; the form in which they were connected to broader programmatic ambitions however persisted: they were still objects of multiplicity and dissension (Firtin, 2023; Weichselberger et al., 2024; Wällstedt, 2020). This means that even though there might be clear formulations and measures regarding the abstract notion of, e.g. “care quality”, activity orchestration could be devised in a variety of ways, and according to different demands.

The process to create a stabilized infrastructure according to Power’s (2015) model took 20 years. Accounting beginnings are apparently hard and arduous work. But once done, it is resilient and continues. From the empirical evidence, there are some clues as to why. Although there was practice stabilization around calculable objects, achieved through the formal articulations in the web-based control system, there was a need to manage the multiplicity of all objects – from quality to competence. This made it important to be flexible (Firtin, 2023; Schrøder et al., 2022). And in order to be flexible, the system had to be strategized by practitioners: it was impossible to work only according to what was specified. Strategizing, we argue, means both working within and outside the boundaries of the formalized system, but always within the bounds of the abstract policy object – the form of the object that harbors a diverse set of programmatic ambitions.

Figure 5 summarizes the findings and extends the framework by Power (2015) in a way that helps us understand how accounting continues. Accounting continues for three interrelated reasons: (1) because of the foundational work that is done over long time, where objects such as quality, competence, fairness and economy go from existing in abstract programmatic forms to more concrete technological ones around which activities can be orchestrated. Here, exercises involving learning and collaboration are key. (2) Because such learning and cooperation allow practitioners – professionals and accountants – to engage in strategizing the system in different ways that make it functional. Such strategizing can be argued to be part of the broader infrastructure, because any object exists in multiple forms – from abstract to concrete at the same time. (3) Because such strategizing ties practice to a diverse set of programmatic ambitions, top managers and politicians come to think that the system works in rather unproblematic ways – they never see the struggles that go on at “the lower levels” of organization.

When accounting continues, the hard and arduous work of accounting beginnings does not have to be repeated: the infrastructure is already in place. We saw that in building it, there was considerable experimentation: creating and producing concrete objects took time. Once the infrastructure is done, any object – from gender equality to collaboration – can be programmatically articulated, elaborated and have activities orchestrated around it within a year or two. As such, it becomes “just another object” to strategize around. In Figure 5 we depict this process as a recurring one. Our concluding argument, here, is that this might be an important part of the puzzle to NPM’s resilience: the way to address anything – any object – is exactly the same; it allows for exactly the same process to be made over and over again. Through strategizing there can be changes between years and political mandate periods, but only on the surface. This is highly convenient for everyone involved: from politicians to accountants and professionals.

In the first chapter of their book, Lapsley and Miller (2024, p. 4) provide a quote from 1983, by an NPM proponent. This proponent depicted the old public administration model, which institutions like, e.g. the NHS was built upon, as “‘a mobile’: designed to move with every breath of air, but which in fact never changes its position and gives no clear indication of direction”. Exactly the same thing can be said about the NPM-related infrastructure we have analyzed. Funck and Karlsson (2023) are obviously right when they argue that proponents of post-NPM movements echo the concerns of NPM proponents “back in the days”. We add to this by arguing that NPM has become what it was supposed to replace: a direction-less mobile, maintaining status quo while allowing for strategic behavior (Caffrey et al., 2019). What we learn from NPM’s early days is that no “radical rethinking” (Parker et al., 2023) or “imaginaries” (Funck and Karlsson, 2023) can change this: what is needed, for a real paradigm shift, is the introduction of new problems, and new experimentation on the practical level. What remains, both from a practical and academic point of view, would be to discuss what constitute “enough experimentation”? What can be introduced today that is as disruptive as the introduction of, e.g. tender documents, purchaser/provider models, profit units, etc., during the implementation of NPM? These questions might have to be answered – otherwise accounting will continue to perform as is.

We would like to thank two anonymous reviewers for their valuable and constructive comments during the review process, which truly helped improve the article. We would also like to thank the participants in the research network MUSICA at Stockholm Business School, Stockholm University, for their valuable comments in our internal research seminars.